feat(mcp_ai_agents): add Strands MCP Agent

[elizabethfuentes12]

Adds a Strands MCP Agent to the MCP AI Agents section.

A Streamlit app that answers questions about the Strands Agents SDK using a Strands agent connected to an MCP server. The Strands MCPClient launches the Strands documentation MCP server locally over stdio (uvx), discovers its tools (search_docs, fetch_doc), and passes them to an agent backed by an OpenAI model (gpt-4o-mini by default).

What this adds

• New folder mcp_ai_agents/strands_mcp_agent/ with strands_mcp_agent.py, README.md, and requirements.txt
• One entry in the root README under MCP AI Agents
• Apache-2.0 (SPDX header in the module)

How it works

1. Starts strands-agents-mcp-server as a local subprocess with uvx
2. The Strands MCPClient connects over stdio and calls list_tools_sync()
3. The discovered tools are passed to a Strands Agent on an OpenAI model
4. The agent calls search_docs and fetch_doc, then answers with citations

Testing

• Syntax check passes and the Strands and MCP imports resolve
• The MCP server connects over stdio and lists its tools (search_docs, fetch_doc)
• The final OpenAI call requires a user-provided API key

Code follows the official Strands MCP client and OpenAI provider docs.

[Shubhamsaboo]

Thanks for this, and for the clean, well-documented code. We're going to pass on this one for now. It's narrow in scope: a docs-Q&A client over the Strands SDK's own MCP server, which is mostly useful to readers already committed to Strands, and as a tutorial it stays fairly close to a canned docs-search demo.

Not a quality concern, the code itself is solid. If you'd like to contribute a broader MCP example (an agent that combines tools or solves a task a reader would reach for beyond querying one SDK's docs), we'd be happy to review it. Closing this for now.

---

Generated by Claude Code